Impacts of climate change on agro-climatic suitability of major food crops and crop diversification potential in Ghana

Crop diversification is a promising climate change adaptation strategy for food production stability. However, without quantitative assessments of where, with which crop mixes and to what extent diversification is possible now and under future climatic conditions, efforts to expand crop diversification under Nationally Determined Contributions (NDCs) and National Action Plans (NAP) are unsystematic. In this study, we used extreme gradient boosting, a machine learning approach to model the current climatic suitability for maize, sorghum, cassava and groundnut in Ghana using yield data and agronomically important variables. We then used multi-model future climate projections for the 2050s and two greenhouse gas emissions scenarios (RCP 2.6 and RCP 8.5) to predict changes in the suitability range of these crops. We achieved a good model fit in determining suitability classes for all crops (AUC=0.81-0.87). Precipitation-based factors are suggested as most important in determining crop suitability, though the importance is crop-specific. Under projected climatic conditions, optimal suitability areas will decrease for all crops except for groundnuts under RCP8.5 (no change: 0%), with greatest losses for maize (12% under RCP2.6 and 14% under RCP8.5). Under current climatic conditions, 18% of Ghana has optimal suitability for two crops, 2% for three crops with no area having optimal suitability for all the four crops. Under projected climatic conditions, areas with optimal suitability for producing two and three crops will decrease by 12% as areas having moderate and marginal conditions for multiple crops increase. We also found that although diversification opportunities are spatially distinct, cassava and groundnut will be more simultaneously suitable for the south while groundnut and sorghum will be more suitable for the northern parts of Ghana under projected climatic conditions.


33
The agricultural sector of tropical countries is at great risk from the impacts of climate change. This is 34 because of changes in weather patterns, which determine yields and crop production in these areas [1,2]. the accuracy expected to occur by chance with k values ranging from -1 (poor) to 1 (good) [30]. The 174 multiclass area under receiver operating characteristic curve (AUC) was used to validate model fit by 175 comparing and averaging all pairwise class AUC. Sensitivity for suitability class is the percentage of a 176 category on the reference data that is correctly modelled as belonging to that category, and measures 177 proportion of pixels omitted from a reference suitability class (omission error). Specificity expresses the 178 proportion of a category on the reference data that is included erroneously in another suitability class 179 (commission error) [69]. Other metrics used for class accuracy are described in full by Galdi and 180 Tagliaferri (70) and Hossin and Sulaiman (71)  In order to determine the diversification potential for the four key food crops for Ghana, we combined the 186 suitability of the crops to understand which areas are suitable for which crops and to what degree. At first 187 the maps were combined to determine the number of crops that were suitable for each cell. To determine 188 suitability for multiple crops, we summed the ranks of the modelled crop suitability with each class ranked 189 from 1 (limited) to 4 (optimal). This produced diversification potential for the four scale from 4 (very low)

190
to 16 (very high). After that, realizing that two-crop combinations were most frequently observed, further To reliably assess crop suitability, we first evaluated the fit of the model on an independent test data set.

200
There were differences in the model fit between crops, but all crops showed a good fit. The best accuracy 201 was for modelling sorghum (OA=0.82, k=0.75 and AUC=0.87) (see Table 3

228
Under current climatic conditions, the suitability of maize is very variable across the country with no 229 specific regional distribution. Optimal suitable areas for maize cover 23% of the country (Figure 3a, 230 Figure 4a). Under projected climatic conditions the areas that have optimal suitability for maize 231 production will decrease by 12% and by 14% under RCP2.6 and RCP8.5 respectively as suitability 232 transition from being optimal to moderately suitable and marginal. These are the largest changes from the 233 optimal suitable areas of the crops modelled in this study. Areas that have limited suitability are projected 234 to increase by 8% under RCP2.6 and by 7% under RCP8.5 scenario, with marginal areas decreasing by 235 11% (RCP82.6) and by 8% (RCP8.5) (Figure 3b-c, Figure 4a and Table 5).

236
Sorghum was modelled as having largest area for which it is optimally suitable (28%), which is the 237 highest of the four crops for this category (Figure 3d, Figure 4b). Under climate change, the optimal 238 suitability areas for sorghum are projected to decrease by 10% and 13% decrease under RCP2.6 and 239 RCP8.5 respectively (Figure 3e-f, Figure 4b). Some parts of northern Ghana that have limited suitability 240 for sorghum will become suitable under both RCP2.6 and RCP8.5 with an evident northwards shift in 241 sorghum suitability under climate change. The areas that are unsuitable for sorghum are projected to 242 increase by 12% (RCP2.6) and 13% (RCP8.5) by the 2050s (Figure 3e-f, Table 5).

243
Cassava was modelled as mostly suitable in the southern forested bimodal rainfall areas of Ghana ( Figure   244 3g). Under RCP2.6, the results show that by the 2050s, optimal suitable areas for cassava will decrease by 245 7% while under RCP28.5, they will decrease by 9% ( Figure 4c). Concurrently, the areas that have limited 246 suitability for cassava will also slightly decrease by 4% under RCP2.6 and by 3% under RCP8.5 from the 247 current 35%. The results showed that 48% of Ghana can produce groundnuts (optimal and moderate that the areas that have optimal suitability for groundnuts will decrease by 3% under RCP8.5 but will 251 remain stable at 17% under RCP8.5, a trajectory different from other crops modelled in this study (Table   252 5, Figure 4d).  RCP8.5, with a concurrent reduction in areas that are moderately suitable for just one crop as these 275 become less (Figure 4b-c, Table 6).

280
We assessed the crop diversification potential under current and projected climatic conditions for pairwise 281 combinations of crops across the country, as results indicated this to be the most promising diversification 282 option ( Figure 6). Areas with a highest potential for both crops in combination are 5.5% for maize and 283 groundnuts, 5.4% for cassava and sorghum and 5.2% for maize and cassava ( Figure 6, Figure 7, SI Table   284 4); all other combinations are below 5%. Except for cassava and groundnut combined, all combinations of 285 crops are projected to decrease for the areas where both crops currently have the high suitability class.

286
Concurrently, the areas where a combination of crops will be moderate and marginal or marginal for both

295
The impacts of climate change on food security are multi-faceted. Therefore, in this study we quantified 296 the impacts of climate change on the crop diversification potential by assessing the suitability of key food 297 crops in the case of Ghana. A model for estimating crop suitability for maize, sorghum, cassava and 298 groundnuts under current climate was constructed, which reliably reproduced observed suitability patterns.

299
Therefore, we deem the model as sufficiently robust to predict the suitability of the four crops under future 300 climate conditions. We identified individual and combined crop suitability for assessing areas with 301 opportunities for crop diversification under current and projected climatic conditions.

389
There are some limitations and potential sources of uncertainty that should be considered in the 390 interpretation of our results. The suitability models are driven by climate data and current crop production 391 data, which have inherent uncertainties. Future projections of crop production suitability are produced by 392 combining suitability models with projections based on GCMs that describe potential future conditions.

393
These different GCMs rely on different parameters and incorporate different functions to cover the 394 dynamics of atmospheric circulation, ocean effects, or feedbacks between the land surface and the 395 atmosphere. Therefore, they are prone to disagreements or errors that will be propagated in the modelling.

396
Our modelling omitted direct physiological interaction effects in multiple crops such as nitrogen fixation,   The 11 year mean crop yield distribution for each of the four suitability classes across all districts. The 695 right axis in (a, b) is for cassava yields only. lines are Gaussian distribution fit for each climatic scenario. Ld-Ld is for limited suitability for both crops, 718 Ld-Mg is area with limited for one of the 2 crops and marginal for the other, Mg-Mg is area where both 719 crops are marginal, Mg-Md is area that is marginal for one crop while moderately suitable for the other 720 crop, Md-Md is where both crops are moderately suitable, Md-Op is where one crop is moderately 721 suitable and the other has optimal suitability and Op-Op is where both crops are optimally suitable. 722